My friend and co-author Lynne Levesque pointed me to a great blog post by Kevin Eikenberry on how to improve brainstorming. I agree with Kevin when he says, "The problem is that we try to brainstorm without a clear problem statement, in the middle of a long meeting, when people are already tired, and they are sitting in a relatively (or totally) sterile, uncreative space." He goes on to offer ten useful tenets for improving how teams brainstorm.
I would add one crucial element to his recommended process. I believe that groups need to do some detective work after they craft a problem statement and before they begin to brainstorm. They need to out into the field, use their powers of observation, and collect data about the phenomenon. That means interacting with, observing, and questioning customers and other constituencies. Once you have gathered some of that fuel for the creative fire, then you can brainstorm effectively. Starting with an absolutely blank slate, in my view, doesn't yield the best results.